If you think that there is an error in how your package is being tested or represented, please file an issue at NewPkgEval.jl , making sure to read the FAQ first.
Results with Julia v1.2.0
Testing was successful .
Last evaluation was ago and took 3 minutes, 16 seconds.
Click here to download the log file.
Click here to show the log contents.
Resolving package versions...
Installed RecipesBase ──────────────────── v0.7.0
Installed Requires ─────────────────────── v0.5.2
Installed LinearTimeVaryingModelsBase ──── v0.2.1
Installed DifferentialDynamicProgramming ─ v0.4.0
Installed ValueHistories ───────────────── v0.5.1
Installed OrderedCollections ───────────── v1.1.0
Installed Parameters ───────────────────── v0.12.0
Installed DataStructures ───────────────── v0.17.6
Updating `~/.julia/environments/v1.2/Project.toml`
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
Updating `~/.julia/environments/v1.2/Manifest.toml`
[864edb3b] + DataStructures v0.17.6
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
[5e7a79f3] + LinearTimeVaryingModelsBase v0.2.1
[bac558e1] + OrderedCollections v1.1.0
[d96e819e] + Parameters v0.12.0
[3cdcf5f2] + RecipesBase v0.7.0
[ae029012] + Requires v0.5.2
[98cad3c8] + ValueHistories v0.5.1
[2a0f44e3] + Base64
[8ba89e20] + Distributed
[b77e0a4c] + InteractiveUtils
[8f399da3] + Libdl
[37e2e46d] + LinearAlgebra
[56ddb016] + Logging
[d6f4376e] + Markdown
[de0858da] + Printf
[9a3f8284] + Random
[9e88b42a] + Serialization
[6462fe0b] + Sockets
[2f01184e] + SparseArrays
[10745b16] + Statistics
[8dfed614] + Test
[4ec0a83e] + Unicode
Testing DifferentialDynamicProgramming
Status `/tmp/jl_Smyce8/Manifest.toml`
[864edb3b] DataStructures v0.17.6
[9cd584b6] DifferentialDynamicProgramming v0.4.0
[5e7a79f3] LinearTimeVaryingModelsBase v0.2.1
[bac558e1] OrderedCollections v1.1.0
[d96e819e] Parameters v0.12.0
[3cdcf5f2] RecipesBase v0.7.0
[ae029012] Requires v0.5.2
[98cad3c8] ValueHistories v0.5.1
[2a0f44e3] Base64 [`@stdlib/Base64`]
[8ba89e20] Distributed [`@stdlib/Distributed`]
[b77e0a4c] InteractiveUtils [`@stdlib/InteractiveUtils`]
[8f399da3] Libdl [`@stdlib/Libdl`]
[37e2e46d] LinearAlgebra [`@stdlib/LinearAlgebra`]
[56ddb016] Logging [`@stdlib/Logging`]
[d6f4376e] Markdown [`@stdlib/Markdown`]
[de0858da] Printf [`@stdlib/Printf`]
[9a3f8284] Random [`@stdlib/Random`]
[9e88b42a] Serialization [`@stdlib/Serialization`]
[6462fe0b] Sockets [`@stdlib/Sockets`]
[2f01184e] SparseArrays [`@stdlib/SparseArrays`]
[10745b16] Statistics [`@stdlib/Statistics`]
[8dfed614] Test [`@stdlib/Test`]
[4ec0a83e] Unicode [`@stdlib/Unicode`]
[ Info: Compile time is high for this package, this is expected and is not an error.
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.1247 17.6 17.6 0.0911 0.0
2 32.5587 12 12 0.0906 -0.2
3 20.5755 8.6 8.6 0.113 -0.6
4 11.9772 4.61 4.61 0.144 -1.2
5 7.36242 1.78 1.78 0.168 -2.0
6 5.58242 0.428 0.428 0.219 -3.1
7 5.15461 0.0223 0.0223 0.0703 -4.3
8 5.13227 2.27e-05 2.27e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.132252
final grad: 3.874959e-06
final λ: 4.4841551e-08
time / iter: 318 ms
total time: 2.86 seconds, of which
derivs: 0.0 %
back pass: 43.1%
fwd pass: 11.6%
other: 45.2% (graphics etc.)
=========== end iLQG ===========
15.669386 seconds (15.73 M allocations: 960.859 MiB, 5.87% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.0885 20.8 20.8 0.107 0.0
2 29.2398 9.11 9.11 0.0838 -0.2
3 20.1308 4.63 4.63 0.082 -0.6
4 15.5039 3.22 3.22 0.118 -1.2
5 12.2853 1.77 1.77 0.214 -2.0
6 10.5119 0.363 0.363 0.211 -3.1
7 10.1489 0.0144 0.0144 0.0593 -4.3
8 10.1345 1.31e-05 1.31e-05 0.00201 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.13451
final grad: 3.248093e-06
final λ: 4.4841551e-08
time / iter: 84 ms
total time: 0.76 seconds, of which
derivs: 0.1 %
back pass: 69.8%
fwd pass: 25.5%
other: 4.6 % (graphics etc.)
=========== end iLQG ===========
0.848369 seconds (1.40 M allocations: 268.639 MiB, 3.90% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0601 21.5 21.5 0.109 0.0
2 29.5361 10.6 10.6 0.0916 -0.2
3 18.8912 6.07 6.07 0.0966 -0.6
4 12.8247 4.26 4.26 0.134 -1.2
5 8.56329 2.68 2.68 0.251 -2.0
6 5.87883 0.587 0.587 0.254 -3.1
7 5.29149 0.0186 0.0186 0.0679 -4.3
8 5.27286 1.28e-05 1.28e-05 0.00229 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.272849
final grad: 3.49079e-06
final λ: 4.4841551e-08
time / iter: 76 ms
total time: 0.68 seconds, of which
derivs: 0.2 %
back pass: 72.0%
fwd pass: 24.7%
other: 3.1 % (graphics etc.)
=========== end iLQG ===========
0.784096 seconds (1.40 M allocations: 268.639 MiB, 3.36% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.2998 20.4 20.4 0.101 0.0
2 29.8672 10.5 10.5 0.0849 -0.2
3 19.4016 7.13 7.13 0.102 -0.6
4 12.2673 4.63 4.63 0.136 -1.2
5 7.64203 2.03 2.03 0.183 -2.0
6 5.61641 0.353 0.353 0.182 -3.1
7 5.26311 0.0122 0.0122 0.0591 -4.3
8 5.25088 9.73e-06 9.73e-06 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.250866
final grad: 3.203902e-06
final λ: 4.4841551e-08
time / iter: 104 ms
total time: 0.94 seconds, of which
derivs: 0.1 %
back pass: 64.3%
fwd pass: 29.1%
other: 6.4 % (graphics etc.)
=========== end iLQG ===========
0.975158 seconds (1.40 M allocations: 268.639 MiB, 14.37% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.7243 26.2 26.2 0.113 0.0
2 23.4751 8.95 8.95 0.0763 -0.2
3 14.5222 3.04 3.04 0.0648 -0.6
4 11.4817 1.14 1.14 0.0717 -1.2
5 10.3406 0.748 0.748 0.13 -2.0
6 9.59252 0.308 0.308 0.184 -3.1
7 9.28464 0.018 0.018 0.055 -4.3
8 9.26669 1.85e-05 1.85e-05 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 9.266675
final grad: 3.309447e-06
final λ: 4.4841551e-08
time / iter: 74 ms
total time: 0.66 seconds, of which
derivs: 0.2 %
back pass: 70.9%
fwd pass: 27.0%
other: 2.0 % (graphics etc.)
=========== end iLQG ===========
0.790482 seconds (1.40 M allocations: 268.639 MiB, 3.51% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.6086 26.8 26.8 0.116 0.0
2 23.799 11.6 11.6 0.0886 -0.2
3 12.1548 4.97 4.97 0.0778 -0.6
4 7.18008 2.4 2.4 0.0833 -1.2
5 4.77537 1.18 1.18 0.135 -2.0
6 3.59532 0.237 0.237 0.167 -3.1
7 3.35866 0.0103 0.0103 0.0556 -4.3
8 3.34832 1.16e-05 1.16e-05 0.00203 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.348307
final grad: 3.361818e-06
final λ: 4.4841551e-08
time / iter: 87 ms
total time: 0.78 seconds, of which
derivs: 0.2 %
back pass: 68.5%
fwd pass: 23.3%
other: 8.0 % (graphics etc.)
=========== end iLQG ===========
0.875774 seconds (1.40 M allocations: 268.639 MiB, 3.00% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 48.8912 22 22 0.103 0.0
2 26.9357 11.1 11.1 0.0851 -0.2
3 15.8772 6.43 6.43 0.0924 -0.6
4 9.45055 3.98 3.98 0.119 -1.2
5 5.46989 2.52 2.52 0.172 -2.0
6 2.95429 0.949 0.949 0.205 -3.1
7 2.0054 0.0605 0.0605 0.0776 -4.3
8 1.94491 6.9e-05 6.9e-05 0.00298 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 1.944839
final grad: 4.641227e-06
final λ: 4.4841551e-08
time / iter: 101 ms
total time: 0.91 seconds, of which
derivs: 0.1 %
back pass: 71.1%
fwd pass: 26.4%
other: 2.4 % (graphics etc.)
=========== end iLQG ===========
0.934042 seconds (1.40 M allocations: 268.639 MiB, 2.60% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.6099 26.7 26.7 0.106 0.0
2 24.8642 11.5 11.5 0.0823 -0.2
3 13.3665 4.88 4.88 0.076 -0.6
4 8.48701 2.25 2.25 0.0841 -1.2
5 6.23922 0.851 0.851 0.143 -2.0
6 5.38792 0.111 0.111 0.132 -3.1
7 5.27669 0.00519 0.00519 0.0456 -4.3
8 5.27149 6.93e-06 6.93e-06 0.00189 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.271487
final grad: 3.213114e-06
final λ: 4.4841551e-08
time / iter: 101 ms
total time: 0.91 seconds, of which
derivs: 0.1 %
back pass: 67.5%
fwd pass: 28.8%
other: 3.6 % (graphics etc.)
=========== end iLQG ===========
1.031367 seconds (1.40 M allocations: 268.639 MiB, 13.73% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0102 26.2 26.2 0.115 0.0
2 24.8497 10.8 10.8 0.085 -0.2
3 14.0186 5.39 5.39 0.0863 -0.6
4 8.62521 3.01 3.01 0.107 -1.2
5 5.61383 1.65 1.65 0.169 -2.0
6 3.96697 0.404 0.404 0.187 -3.1
7 3.56312 0.018 0.018 0.0558 -4.3
8 3.54514 1.79e-05 1.79e-05 0.00216 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.545121
final grad: 3.581447e-06
final λ: 4.4841551e-08
time / iter: 86 ms
total time: 0.78 seconds, of which
derivs: 0.1 %
back pass: 74.1%
fwd pass: 22.4%
other: 3.3 % (graphics etc.)
=========== end iLQG ===========
0.869140 seconds (1.40 M allocations: 268.639 MiB, 3.27% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.642 15.2 15.2 0.0855 0.0
2 35.3962 9.74 9.74 0.0823 -0.2
3 25.6515 7.09 7.09 0.105 -0.6
4 18.5648 4.8 4.8 0.16 -1.2
5 13.7694 2.4 2.4 0.227 -2.0
6 11.372 0.627 0.627 0.233 -3.1
7 10.745 0.0337 0.0337 0.0694 -4.3
8 10.7113 4.03e-05 4.03e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.71122
final grad: 3.964869e-06
final λ: 4.4841551e-08
time / iter: 86 ms
total time: 0.77 seconds, of which
derivs: 0.1 %
back pass: 68.1%
fwd pass: 26.0%
other: 5.8 % (graphics etc.)
=========== end iLQG ===========
0.821629 seconds (1.40 M allocations: 268.640 MiB, 3.18% gc time)
Running linear demo function for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.3316 13.8 13.8 0.0813 0.0
2 35.5392 8.61 8.61 0.0773 -0.2
3 26.9284 6.45 6.45 0.099 -0.6
4 20.4758 4.1 4.1 0.149 -1.2
5 16.3735 1.5 1.5 0.196 -2.0
6 14.8783 0.223 0.223 0.163 -3.1
7 14.6555 0.0077 0.0077 0.0513 -4.3
8 14.6478 7.76e-06 7.76e-06 0.00194 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 14.64778
final grad: 3.237175e-06
final λ: 4.4841551e-08
time / iter: 105 ms
total time: 0.95 seconds, of which
derivs: 0.1 %
back pass: 77.4%
fwd pass: 18.9%
other: 3.6 % (graphics etc.)
=========== end iLQG ===========
5.506450 seconds (3.29 M allocations: 361.836 MiB, 5.01% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Running linear demo function with KL-divergence constraint for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 22.3322 26 17.3 0.123 -1.00 0.0095 1.83
2 1.28069e+10 -1.28e+10 5.37e+18 1.41e+07 -0.15 1.28e+10 -1.42
3 4.42424e+07 -4.42e+07 2.1e+15 5.75e+05 -0.07 4.39e+07 0.248
4 225219 -2.25e+05 9.57e+11 1.95e+04 -0.04 2.17e+05 1.07
5 1184.11 -1.14e+03 1.26e+09 945 -0.02 975 1.47
6 10.5919 37.7 7.69e+06 78.6 -0.03 1.46 1.67
7 123.307 -75 1.29e+08 312 -0.03 80.1 1.57
8 417.204 -369 4.22e+08 557 -0.03 317 1.52
9 232.39 -184 2.36e+08 420 -0.03 166 1.55
10 170.332 -122 1.75e+08 363 -0.03 116 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 145.148 -96.8 1.5e+08 337 -0.03 96.9 1.57
SUCCESS: abs(KL-divergence) < kl_step
iterations: 11
final cost: 145.1481
final grad: 336.8931
final λ: 9.3750325e-01
time / iter: 271 ms
total time: 2.98 seconds, of which
derivs: 0.0 %
back pass: 68.1%
fwd pass: 20.2%
other: 11.6% (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 145.1481073547954
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 145.037 0.111 0.0553 0.000151 -1.00 5.58e-05 1.57
2 2.90811e+55 -2.91e+55 1.05e+60 5.18e+25 -0.15 2.92e+55 -1.16
3 1.15096e+52 -1.15e+52 1.9e+56 2.01e+24 -0.07 1.17e+52 0.501
4 3.37113e+43 -3.37e+43 2.88e+47 2.66e+20 -0.04 4.69e+43 1.28
5 4.71286e+11 -4.71e+11 4.77e+15 2.03e+05 -0.02 4.61e+12 1.55
6 145.033 0.115 0.095 0.00299 -0.03 9.61e-05 1.57
7 145.657 -0.509 584 0.574 -0.03 0.564 1.56
8 12564.7 -1.24e+04 1.85e+08 115 -0.03 1.82e+05 1.56
9 138.563 6.59 1.43e+05 6.26 -0.03 142 1.56
10 145.127 0.0214 6.43e+03 1.76 -0.03 6.33 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 143.3 1.85 2.83e+04 3.27 -0.03 28 1.56
EXIT: η > ηmax
iterations: 11
final cost: 143.2996
final grad: 3.267209
final λ: 9.3442150e-01
time / iter: 141 ms
total time: 1.55 seconds, of which
derivs: 0.0 %
back pass: 58.8%
fwd pass: 37.3%
other: 3.8 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 143.2996378785932
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 143.191 0.108 0.0541 0.000122 -1.00 5.46e-05 1.56
2 1.55674e+56 -1.56e+56 7.17e+60 1.68e+26 -0.15 1.57e+56 -1.15
3 1.68786e+52 -1.69e+52 2.71e+56 2.71e+24 -0.07 1.72e+52 0.505
4 3.1857e+43 -3.19e+43 2.94e+47 2.96e+20 -0.04 4.49e+43 1.28
5 1.39165e+11 -1.39e+11 9.75e+14 8.81e+04 -0.02 9.38e+11 1.55
6 143.187 0.113 0.0787 0.00117 -0.03 7.96e-05 1.56
7 143.893 -0.593 27.3 0.12 -0.03 0.0269 1.56
8 1666.27 -1.52e+03 2.06e+07 39.2 -0.03 2.03e+04 1.55
9 143.946 -0.647 8e+03 1.58 -0.03 7.93 1.56
10 145.28 -1.98 3.64e+05 7.52 -0.03 359 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 142.342 0.958 5.21e+04 3.4 -0.03 51.6 1.56
EXIT: η > ηmax
iterations: 11
final cost: 142.3417
final grad: 3.39533
final λ: 9.3196340e-01
time / iter: 129 ms
total time: 1.42 seconds, of which
derivs: 0.0 %
back pass: 52.4%
fwd pass: 40.4%
other: 7.2 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 142.3416956113604
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 142.237 0.104 0.0522 0.000152 -1.00 5.26e-05 1.56
2 4.33473e+55 -4.33e+55 2.03e+60 1.08e+26 -0.15 4.37e+55 -1.15
3 2.29364e+52 -2.29e+52 3.65e+56 4.75e+24 -0.07 2.33e+52 0.507
4 2.92665e+43 -2.93e+43 2.9e+47 3.69e+20 -0.04 4.18e+43 1.28
5 4.61146e+10 -4.61e+10 2.91e+14 4.56e+04 -0.02 2.79e+11 1.55
6 142.23 0.112 0.0977 0.0015 -0.03 9.85e-05 1.56
7 142.844 -0.502 13.9 0.0458 -0.03 0.0138 1.55
8 467.82 -325 3.21e+06 15.5 -0.03 3.18e+03 1.55
9 144.594 -2.25 813 0.521 -0.03 0.807 1.55
10 150.678 -8.34 4.06e+04 2.68 -0.03 40.2 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 179.865 -37.5 3.5e+05 6.39 -0.03 347 1.55
EXIT: η > ηmax
┌ Warning: KL divergence too high for some time steps when done
└ @ DifferentialDynamicProgramming ~/.julia/packages/DifferentialDynamicProgramming/DG1NV/src/iLQGkl.jl:248
iterations: 11
final cost: 179.865
final grad: 6.38689
final λ: 9.3073677e-01
time / iter: 134 ms
total time: 1.47 seconds, of which
derivs: 0.0 %
back pass: 49.4%
fwd pass: 45.2%
other: 5.5 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 179.8649740066802
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 179.749 0.116 0.0578 0.000165 -1.00 5.82e-05 1.55
2 4.79408e+56 -4.79e+56 1.98e+61 2.05e+26 -0.15 4.84e+56 -1.15
3 3.109e+52 -3.11e+52 4.99e+56 4.39e+24 -0.07 3.17e+52 0.509
4 2.60718e+43 -2.61e+43 2.8e+47 3.42e+20 -0.04 3.77e+43 1.29
5 1.48368e+10 -1.48e+10 9.3e+13 2.28e+04 -0.02 8.9e+10 1.54
6 179.709 0.156 0.33 0.00264 -0.03 0.00033 1.55
7 179.366 0.499 55.2 0.0431 -0.03 0.0549 1.55
8 363.425 -184 7.17e+05 5.69 -0.03 708 1.55
9 180.244 -0.379 530 0.222 -0.03 0.526 1.55
10 190.819 -11 8.04e+03 0.95 -0.03 7.96 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 220.85 -41 7.1e+04 2.28 -0.03 70.2 1.55
EXIT: η > ηmax
iterations: 11
final cost: 220.8503
final grad: 2.281649
final λ: 9.2951176e-01
time / iter: 125 ms
total time: 1.37 seconds, of which
derivs: 0.0 %
back pass: 51.1%
fwd pass: 41.5%
other: 7.4 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 220.85032087779095
18.370648 seconds (25.04 M allocations: 3.631 GiB, 6.63% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Entering iLQG function
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 33409.7 2.21e+03 3.57e+03 2.35 0.0
2 31195 705 1.57e+03 0.7 -0.2
3 30489.9 537 1.82e+03 0.719 -0.6
4 29952.6 13.9 479 0.406 -1.2
5 29938.7 224 896 0.427 -2.0
6 29714.5 257 260 0.254 -3.1
7 29457.7 10.2 86.3 0.159 -4.3
8 29447.5 79.8 90.3 0.16 -5.7
9 29367.7 18.1 20.4 0.0844 -7.3
10 29349.5 4.78 5.23 0.0469 -9.2
iteration cost reduction expected gradient log10(λ) η divergence
11 29344.8 0.633 1.81 0.027 -11.2
12 29344.1 1.25 1.62 0.0218 -13.5
13 29342.9 1.38 2.12 0.0204 -15.9
14 29341.5 0.221 0.471 0.0109 -18.6
15 29341.3 0.0322 0.111 0.00601 -21.4
16 29341.2 0.0121 0.0284 0.00338 -24.5
17 29341.2 0.0289 0.00881 0.00194 -27.8
18 29341.2 0.146 0.00967 0.00162 -31.2
19 NO STEP -8.15e-05 4.49e-05 0.00151 -6.0
20 NO STEP -8.15e-05 4.49e-05 0.00151 -5.8
iteration cost reduction expected gradient log10(λ) η divergence
21 NO STEP -8.15e-05 4.49e-05 0.00151 -5.4
22 NO STEP -8.15e-05 4.49e-05 0.00151 -4.8
23 NO STEP -8.15e-05 4.49e-05 0.00151 -4.0
24 NO STEP -8.15e-05 4.49e-05 0.00151 -2.9
25 NO STEP -8.15e-05 4.49e-05 0.00151 -1.7
26 NO STEP -8.15e-05 4.49e-05 0.00151 -0.3
27 NO STEP -8.1e-05 4.49e-05 0.00151 1.3
28 NO STEP -6.37e-05 4.28e-05 0.00142 3.2
29 29341 0.0181 0.00494 0.000294 3.2
30 29341 0.0108 0.00163 0.000587 3.0
iteration cost reduction expected gradient log10(λ) η divergence
31 29341 0.0124 0.00142 0.000705 2.6
32 NO STEP -1.52e-05 4.11e-06 0.000713 1.3
33 NO STEP -3.25e-05 8.4e-06 0.00108 1.6
34 NO STEP -2.93e-05 7.42e-06 0.000994 2.0
35 NO STEP -1.52e-05 4.11e-06 0.000713 2.6
36 29341 0.000673 0.00041 0.000329 2.6
37 NO STEP -4.48e-06 1.01e-06 0.000377 2.2
38 NO STEP -1.02e-05 1.94e-06 0.000507 2.4
39 NO STEP -4.48e-06 1.01e-06 0.000377 2.8
40 NO STEP -6.87e-07 2.39e-07 0.000187 3.4
iteration cost reduction expected gradient log10(λ) η divergence
41 NO STEP -7.03e-08 2.96e-08 5.16e-05 4.2
42 NO STEP -1.2e-09 2.86e-09 8.09e-06 5.2
43 29341 5.78e-07 3.77e-07 7.76e-07 5.2
44 29341 5.64e-07 6e-07 1.24e-06 5.0
45 NO STEP -2.04e-10 1.02e-09 3.16e-06 4.2
46 NO STEP -2.87e-09 2.84e-09 8.08e-06 4.4
47 NO STEP -9.53e-10 1.69e-09 5.06e-06 4.8
48 29341 1.43e-07 9.56e-07 1.98e-06 4.8
49 NO STEP -7.64e-10 1.01e-09 3.16e-06 4.4
50 NO STEP -1.78e-09 1.68e-09 5.05e-06 4.6
iteration cost reduction expected gradient log10(λ) η divergence
51 NO STEP -7.64e-10 1.01e-09 3.16e-06 5.0
52 NO STEP -1.2e-10 3.82e-10 1.23e-06 5.6
53 29341 4e-11 3.99e-10 3.01e-07 5.6
SUCCESS: cost change < tol_fun
iterations: 53
final cost: 29341
final grad: 3.014034e-07
final λ: 2.6959947e+05
time / iter: 648 ms
total time: 34.33 seconds, of which
derivs: 50.8%
back pass: 33.7%
fwd pass: 14.7%
other: 0.7 % (graphics etc.)
=========== end iLQG ===========
Install package Plots.jl (and call using Plots) to plot results in the end of demo_pendcart
Done
==========
Starting box-QP, dimension 500, initial value: 66427.406
RESULT: Gradient norm smaller than tolerance.
iterations 15 gradient 2.39487e-12 final value -31.2095 factorizations 14
2.096864 seconds (1.95 k allocations: 78.105 MiB, 17.51% gc time)
Testing DifferentialDynamicProgramming tests passed
Results with Julia v1.3.0
Testing was successful .
Last evaluation was ago and took 3 minutes, 15 seconds.
Click here to download the log file.
Click here to show the log contents.
Resolving package versions...
Installed Requires ─────────────────────── v0.5.2
Installed RecipesBase ──────────────────── v0.7.0
Installed LinearTimeVaryingModelsBase ──── v0.2.1
Installed DifferentialDynamicProgramming ─ v0.4.0
Installed ValueHistories ───────────────── v0.5.1
Installed OrderedCollections ───────────── v1.1.0
Installed Parameters ───────────────────── v0.12.0
Installed DataStructures ───────────────── v0.17.6
Updating `~/.julia/environments/v1.3/Project.toml`
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
Updating `~/.julia/environments/v1.3/Manifest.toml`
[864edb3b] + DataStructures v0.17.6
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
[5e7a79f3] + LinearTimeVaryingModelsBase v0.2.1
[bac558e1] + OrderedCollections v1.1.0
[d96e819e] + Parameters v0.12.0
[3cdcf5f2] + RecipesBase v0.7.0
[ae029012] + Requires v0.5.2
[98cad3c8] + ValueHistories v0.5.1
[2a0f44e3] + Base64
[8ba89e20] + Distributed
[b77e0a4c] + InteractiveUtils
[8f399da3] + Libdl
[37e2e46d] + LinearAlgebra
[56ddb016] + Logging
[d6f4376e] + Markdown
[de0858da] + Printf
[9a3f8284] + Random
[9e88b42a] + Serialization
[6462fe0b] + Sockets
[2f01184e] + SparseArrays
[10745b16] + Statistics
[8dfed614] + Test
[4ec0a83e] + Unicode
Testing DifferentialDynamicProgramming
Status `/tmp/jl_vmLzz6/Manifest.toml`
[864edb3b] DataStructures v0.17.6
[9cd584b6] DifferentialDynamicProgramming v0.4.0
[5e7a79f3] LinearTimeVaryingModelsBase v0.2.1
[bac558e1] OrderedCollections v1.1.0
[d96e819e] Parameters v0.12.0
[3cdcf5f2] RecipesBase v0.7.0
[ae029012] Requires v0.5.2
[98cad3c8] ValueHistories v0.5.1
[2a0f44e3] Base64 [`@stdlib/Base64`]
[8ba89e20] Distributed [`@stdlib/Distributed`]
[b77e0a4c] InteractiveUtils [`@stdlib/InteractiveUtils`]
[8f399da3] Libdl [`@stdlib/Libdl`]
[37e2e46d] LinearAlgebra [`@stdlib/LinearAlgebra`]
[56ddb016] Logging [`@stdlib/Logging`]
[d6f4376e] Markdown [`@stdlib/Markdown`]
[de0858da] Printf [`@stdlib/Printf`]
[9a3f8284] Random [`@stdlib/Random`]
[9e88b42a] Serialization [`@stdlib/Serialization`]
[6462fe0b] Sockets [`@stdlib/Sockets`]
[2f01184e] SparseArrays [`@stdlib/SparseArrays`]
[10745b16] Statistics [`@stdlib/Statistics`]
[8dfed614] Test [`@stdlib/Test`]
[4ec0a83e] Unicode [`@stdlib/Unicode`]
[ Info: Compile time is high for this package, this is expected and is not an error.
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.1247 17.6 17.6 0.0911 0.0
2 32.5587 12 12 0.0906 -0.2
3 20.5755 8.6 8.6 0.113 -0.6
4 11.9772 4.61 4.61 0.144 -1.2
5 7.36242 1.78 1.78 0.168 -2.0
6 5.58242 0.428 0.428 0.219 -3.1
7 5.15461 0.0223 0.0223 0.0703 -4.3
8 5.13227 2.27e-05 2.27e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.132252
final grad: 3.874959e-06
final λ: 4.4841551e-08
time / iter: 278 ms
total time: 2.50 seconds, of which
derivs: 0.0 %
back pass: 35.8%
fwd pass: 8.1 %
other: 56.1% (graphics etc.)
=========== end iLQG ===========
18.330153 seconds (16.51 M allocations: 992.102 MiB, 6.24% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.0885 20.8 20.8 0.107 0.0
2 29.2398 9.11 9.11 0.0838 -0.2
3 20.1308 4.63 4.63 0.082 -0.6
4 15.5039 3.22 3.22 0.118 -1.2
5 12.2853 1.77 1.77 0.214 -2.0
6 10.5119 0.363 0.363 0.211 -3.1
7 10.1489 0.0144 0.0144 0.0593 -4.3
8 10.1345 1.31e-05 1.31e-05 0.00201 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.13451
final grad: 3.248093e-06
final λ: 4.4841551e-08
time / iter: 70 ms
total time: 0.63 seconds, of which
derivs: 0.2 %
back pass: 66.1%
fwd pass: 27.9%
other: 5.8 % (graphics etc.)
=========== end iLQG ===========
0.739430 seconds (1.35 M allocations: 266.722 MiB, 5.67% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0601 21.5 21.5 0.109 0.0
2 29.5361 10.6 10.6 0.0916 -0.2
3 18.8912 6.07 6.07 0.0966 -0.6
4 12.8247 4.26 4.26 0.134 -1.2
5 8.56329 2.68 2.68 0.251 -2.0
6 5.87883 0.587 0.587 0.254 -3.1
7 5.29149 0.0186 0.0186 0.0679 -4.3
8 5.27286 1.28e-05 1.28e-05 0.00229 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.272849
final grad: 3.49079e-06
final λ: 4.4841551e-08
time / iter: 62 ms
total time: 0.56 seconds, of which
derivs: 0.2 %
back pass: 64.5%
fwd pass: 32.5%
other: 2.8 % (graphics etc.)
=========== end iLQG ===========
0.661148 seconds (1.35 M allocations: 266.722 MiB, 5.01% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.2998 20.4 20.4 0.101 0.0
2 29.8672 10.5 10.5 0.0849 -0.2
3 19.4016 7.13 7.13 0.102 -0.6
4 12.2673 4.63 4.63 0.136 -1.2
5 7.64203 2.03 2.03 0.183 -2.0
6 5.61641 0.353 0.353 0.182 -3.1
7 5.26311 0.0122 0.0122 0.0591 -4.3
8 5.25088 9.73e-06 9.73e-06 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.250866
final grad: 3.203902e-06
final λ: 4.4841551e-08
time / iter: 71 ms
total time: 0.64 seconds, of which
derivs: 0.2 %
back pass: 70.9%
fwd pass: 26.8%
other: 2.2 % (graphics etc.)
=========== end iLQG ===========
0.737823 seconds (1.35 M allocations: 266.722 MiB, 4.04% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.7243 26.2 26.2 0.113 0.0
2 23.4751 8.95 8.95 0.0763 -0.2
3 14.5222 3.04 3.04 0.0648 -0.6
4 11.4817 1.14 1.14 0.0717 -1.2
5 10.3406 0.748 0.748 0.13 -2.0
6 9.59252 0.308 0.308 0.184 -3.1
7 9.28464 0.018 0.018 0.055 -4.3
8 9.26669 1.85e-05 1.85e-05 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 9.266675
final grad: 3.309447e-06
final λ: 4.4841551e-08
time / iter: 84 ms
total time: 0.75 seconds, of which
derivs: 0.2 %
back pass: 65.9%
fwd pass: 26.5%
other: 7.4 % (graphics etc.)
=========== end iLQG ===========
0.860227 seconds (1.35 M allocations: 266.723 MiB, 4.09% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.6086 26.8 26.8 0.116 0.0
2 23.799 11.6 11.6 0.0886 -0.2
3 12.1548 4.97 4.97 0.0778 -0.6
4 7.18008 2.4 2.4 0.0833 -1.2
5 4.77537 1.18 1.18 0.135 -2.0
6 3.59532 0.237 0.237 0.167 -3.1
7 3.35866 0.0103 0.0103 0.0556 -4.3
8 3.34832 1.16e-05 1.16e-05 0.00203 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.348307
final grad: 3.361818e-06
final λ: 4.4841551e-08
time / iter: 86 ms
total time: 0.77 seconds, of which
derivs: 0.2 %
back pass: 67.3%
fwd pass: 31.5%
other: 1.0 % (graphics etc.)
=========== end iLQG ===========
0.790986 seconds (1.35 M allocations: 266.723 MiB, 4.12% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 48.8912 22 22 0.103 0.0
2 26.9357 11.1 11.1 0.0851 -0.2
3 15.8772 6.43 6.43 0.0924 -0.6
4 9.45055 3.98 3.98 0.119 -1.2
5 5.46989 2.52 2.52 0.172 -2.0
6 2.95429 0.949 0.949 0.205 -3.1
7 2.0054 0.0605 0.0605 0.0776 -4.3
8 1.94491 6.9e-05 6.9e-05 0.00298 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 1.944839
final grad: 4.641227e-06
final λ: 4.4841551e-08
time / iter: 82 ms
total time: 0.74 seconds, of which
derivs: 0.1 %
back pass: 67.2%
fwd pass: 29.8%
other: 2.9 % (graphics etc.)
=========== end iLQG ===========
0.763453 seconds (1.35 M allocations: 266.722 MiB, 4.86% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.6099 26.7 26.7 0.106 0.0
2 24.8642 11.5 11.5 0.0823 -0.2
3 13.3665 4.88 4.88 0.076 -0.6
4 8.48701 2.25 2.25 0.0841 -1.2
5 6.23922 0.851 0.851 0.143 -2.0
6 5.38792 0.111 0.111 0.132 -3.1
7 5.27669 0.00519 0.00519 0.0456 -4.3
8 5.27149 6.93e-06 6.93e-06 0.00189 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.271487
final grad: 3.213114e-06
final λ: 4.4841551e-08
time / iter: 80 ms
total time: 0.72 seconds, of which
derivs: 0.2 %
back pass: 65.2%
fwd pass: 29.4%
other: 5.3 % (graphics etc.)
=========== end iLQG ===========
0.836000 seconds (1.35 M allocations: 266.722 MiB, 5.00% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0102 26.2 26.2 0.115 0.0
2 24.8497 10.8 10.8 0.085 -0.2
3 14.0186 5.39 5.39 0.0863 -0.6
4 8.62521 3.01 3.01 0.107 -1.2
5 5.61383 1.65 1.65 0.169 -2.0
6 3.96697 0.404 0.404 0.187 -3.1
7 3.56312 0.018 0.018 0.0558 -4.3
8 3.54514 1.79e-05 1.79e-05 0.00216 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.545121
final grad: 3.581447e-06
final λ: 4.4841551e-08
time / iter: 90 ms
total time: 0.81 seconds, of which
derivs: 0.1 %
back pass: 67.0%
fwd pass: 26.3%
other: 6.5 % (graphics etc.)
=========== end iLQG ===========
0.893927 seconds (1.35 M allocations: 266.722 MiB, 3.41% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.642 15.2 15.2 0.0855 0.0
2 35.3962 9.74 9.74 0.0823 -0.2
3 25.6515 7.09 7.09 0.105 -0.6
4 18.5648 4.8 4.8 0.16 -1.2
5 13.7694 2.4 2.4 0.227 -2.0
6 11.372 0.627 0.627 0.233 -3.1
7 10.745 0.0337 0.0337 0.0694 -4.3
8 10.7113 4.03e-05 4.03e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.71122
final grad: 3.964869e-06
final λ: 4.4841551e-08
time / iter: 92 ms
total time: 0.83 seconds, of which
derivs: 0.1 %
back pass: 67.1%
fwd pass: 28.9%
other: 3.9 % (graphics etc.)
=========== end iLQG ===========
0.925468 seconds (1.35 M allocations: 266.723 MiB, 3.25% gc time)
Running linear demo function for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.3316 13.8 13.8 0.0813 0.0
2 35.5392 8.61 8.61 0.0773 -0.2
3 26.9284 6.45 6.45 0.099 -0.6
4 20.4758 4.1 4.1 0.149 -1.2
5 16.3735 1.5 1.5 0.196 -2.0
6 14.8783 0.223 0.223 0.163 -3.1
7 14.6555 0.0077 0.0077 0.0513 -4.3
8 14.6478 7.76e-06 7.76e-06 0.00194 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 14.64778
final grad: 3.237175e-06
final λ: 4.4841551e-08
time / iter: 124 ms
total time: 1.11 seconds, of which
derivs: 0.1 %
back pass: 72.0%
fwd pass: 26.1%
other: 1.8 % (graphics etc.)
=========== end iLQG ===========
5.910496 seconds (3.12 M allocations: 353.599 MiB, 7.44% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Running linear demo function with KL-divergence constraint for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 22.3322 26 17.3 0.123 -1.00 0.0095 1.83
2 1.28069e+10 -1.28e+10 5.37e+18 1.41e+07 -0.15 1.28e+10 -1.42
3 4.42424e+07 -4.42e+07 2.1e+15 5.75e+05 -0.07 4.39e+07 0.248
4 225219 -2.25e+05 9.57e+11 1.95e+04 -0.04 2.17e+05 1.07
5 1184.11 -1.14e+03 1.26e+09 945 -0.02 975 1.47
6 10.5919 37.7 7.69e+06 78.6 -0.03 1.46 1.67
7 123.307 -75 1.29e+08 312 -0.03 80.1 1.57
8 417.204 -369 4.22e+08 557 -0.03 317 1.52
9 232.39 -184 2.36e+08 420 -0.03 166 1.55
10 170.332 -122 1.75e+08 363 -0.03 116 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 145.148 -96.8 1.5e+08 337 -0.03 96.9 1.57
SUCCESS: abs(KL-divergence) < kl_step
iterations: 11
final cost: 145.1481
final grad: 336.8931
final λ: 9.3750325e-01
time / iter: 297 ms
total time: 3.27 seconds, of which
derivs: 0.0 %
back pass: 68.5%
fwd pass: 22.1%
other: 9.4 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 145.1481073547954
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 145.037 0.111 0.0553 0.000151 -1.00 5.58e-05 1.57
2 2.90811e+55 -2.91e+55 1.05e+60 5.18e+25 -0.15 2.92e+55 -1.16
3 1.15096e+52 -1.15e+52 1.9e+56 2.01e+24 -0.07 1.17e+52 0.501
4 3.37113e+43 -3.37e+43 2.88e+47 2.66e+20 -0.04 4.69e+43 1.28
5 4.71286e+11 -4.71e+11 4.77e+15 2.03e+05 -0.02 4.61e+12 1.55
6 145.033 0.115 0.095 0.00299 -0.03 9.61e-05 1.57
7 145.657 -0.509 584 0.574 -0.03 0.564 1.56
8 12564.7 -1.24e+04 1.85e+08 115 -0.03 1.82e+05 1.56
9 138.563 6.59 1.43e+05 6.26 -0.03 142 1.56
10 145.127 0.0214 6.43e+03 1.76 -0.03 6.33 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 143.3 1.85 2.83e+04 3.27 -0.03 28 1.56
EXIT: η > ηmax
iterations: 11
final cost: 143.2996
final grad: 3.267209
final λ: 9.3442150e-01
time / iter: 160 ms
total time: 1.76 seconds, of which
derivs: 0.0 %
back pass: 55.1%
fwd pass: 38.3%
other: 6.6 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 143.2996378785932
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 143.191 0.108 0.0541 0.000122 -1.00 5.46e-05 1.56
2 1.55674e+56 -1.56e+56 7.17e+60 1.68e+26 -0.15 1.57e+56 -1.15
3 1.68786e+52 -1.69e+52 2.71e+56 2.71e+24 -0.07 1.72e+52 0.505
4 3.1857e+43 -3.19e+43 2.94e+47 2.96e+20 -0.04 4.49e+43 1.28
5 1.39165e+11 -1.39e+11 9.75e+14 8.81e+04 -0.02 9.38e+11 1.55
6 143.187 0.113 0.0787 0.00117 -0.03 7.96e-05 1.56
7 143.893 -0.593 27.3 0.12 -0.03 0.0269 1.56
8 1666.27 -1.52e+03 2.06e+07 39.2 -0.03 2.03e+04 1.55
9 143.946 -0.647 8e+03 1.58 -0.03 7.93 1.56
10 145.28 -1.98 3.64e+05 7.52 -0.03 359 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 142.342 0.958 5.21e+04 3.4 -0.03 51.6 1.56
EXIT: η > ηmax
iterations: 11
final cost: 142.3417
final grad: 3.39533
final λ: 9.3196340e-01
time / iter: 156 ms
total time: 1.72 seconds, of which
derivs: 0.0 %
back pass: 60.1%
fwd pass: 36.6%
other: 3.2 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 142.3416956113604
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 142.237 0.104 0.0522 0.000152 -1.00 5.26e-05 1.56
2 4.33473e+55 -4.33e+55 2.03e+60 1.08e+26 -0.15 4.37e+55 -1.15
3 2.29364e+52 -2.29e+52 3.65e+56 4.75e+24 -0.07 2.33e+52 0.507
4 2.92665e+43 -2.93e+43 2.9e+47 3.69e+20 -0.04 4.18e+43 1.28
5 4.61146e+10 -4.61e+10 2.91e+14 4.56e+04 -0.02 2.79e+11 1.55
6 142.23 0.112 0.0977 0.0015 -0.03 9.85e-05 1.56
7 142.844 -0.502 13.9 0.0458 -0.03 0.0138 1.55
8 467.82 -325 3.21e+06 15.5 -0.03 3.18e+03 1.55
9 144.594 -2.25 813 0.521 -0.03 0.807 1.55
10 150.678 -8.34 4.06e+04 2.68 -0.03 40.2 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 179.865 -37.5 3.5e+05 6.39 -0.03 347 1.55
EXIT: η > ηmax
┌ Warning: KL divergence too high for some time steps when done
└ @ DifferentialDynamicProgramming ~/.julia/packages/DifferentialDynamicProgramming/DG1NV/src/iLQGkl.jl:248
iterations: 11
final cost: 179.865
final grad: 6.38689
final λ: 9.3073677e-01
time / iter: 126 ms
total time: 1.39 seconds, of which
derivs: 0.0 %
back pass: 53.7%
fwd pass: 42.0%
other: 4.4 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 179.8649740066802
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 179.749 0.116 0.0578 0.000165 -1.00 5.82e-05 1.55
2 4.79408e+56 -4.79e+56 1.98e+61 2.05e+26 -0.15 4.84e+56 -1.15
3 3.109e+52 -3.11e+52 4.99e+56 4.39e+24 -0.07 3.17e+52 0.509
4 2.60718e+43 -2.61e+43 2.8e+47 3.42e+20 -0.04 3.77e+43 1.29
5 1.48368e+10 -1.48e+10 9.3e+13 2.28e+04 -0.02 8.9e+10 1.54
6 179.709 0.156 0.33 0.00264 -0.03 0.00033 1.55
7 179.366 0.499 55.2 0.0431 -0.03 0.0549 1.55
8 363.425 -184 7.17e+05 5.69 -0.03 708 1.55
9 180.244 -0.379 530 0.222 -0.03 0.526 1.55
10 190.819 -11 8.04e+03 0.95 -0.03 7.96 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 220.85 -41 7.1e+04 2.28 -0.03 70.2 1.55
EXIT: η > ηmax
iterations: 11
final cost: 220.8503
final grad: 2.281649
final λ: 9.2951176e-01
time / iter: 127 ms
total time: 1.39 seconds, of which
derivs: 0.0 %
back pass: 54.0%
fwd pass: 43.6%
other: 2.4 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 220.85032087779095
17.762983 seconds (22.20 M allocations: 3.498 GiB, 6.29% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Entering iLQG function
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 33409.7 2.21e+03 3.57e+03 2.35 0.0
2 31195 705 1.57e+03 0.7 -0.2
3 30489.9 537 1.82e+03 0.719 -0.6
4 29952.6 13.9 479 0.406 -1.2
5 29938.7 224 896 0.427 -2.0
6 29714.5 257 260 0.254 -3.1
7 29457.7 10.2 86.3 0.159 -4.3
8 29447.5 79.8 90.3 0.16 -5.7
9 29367.7 18.1 20.4 0.0844 -7.3
10 29349.5 4.78 5.23 0.0469 -9.2
iteration cost reduction expected gradient log10(λ) η divergence
11 29344.8 0.633 1.81 0.027 -11.2
12 29344.1 1.25 1.62 0.0218 -13.5
13 29342.9 1.38 2.12 0.0204 -15.9
14 29341.5 0.221 0.471 0.0109 -18.6
15 29341.3 0.0322 0.111 0.00601 -21.4
16 29341.2 0.0121 0.0284 0.00338 -24.5
17 29341.2 0.0289 0.00881 0.00194 -27.8
18 29341.2 0.146 0.00967 0.00162 -31.2
19 NO STEP -8.15e-05 4.49e-05 0.00151 -6.0
20 NO STEP -8.15e-05 4.49e-05 0.00151 -5.8
iteration cost reduction expected gradient log10(λ) η divergence
21 NO STEP -8.15e-05 4.49e-05 0.00151 -5.4
22 NO STEP -8.15e-05 4.49e-05 0.00151 -4.8
23 NO STEP -8.15e-05 4.49e-05 0.00151 -4.0
24 NO STEP -8.15e-05 4.49e-05 0.00151 -2.9
25 NO STEP -8.15e-05 4.49e-05 0.00151 -1.7
26 NO STEP -8.15e-05 4.49e-05 0.00151 -0.3
27 NO STEP -8.1e-05 4.49e-05 0.00151 1.3
28 NO STEP -6.37e-05 4.28e-05 0.00142 3.2
29 29341 0.0181 0.00494 0.000294 3.2
30 29341 0.0108 0.00163 0.000587 3.0
iteration cost reduction expected gradient log10(λ) η divergence
31 29341 0.0124 0.00142 0.000705 2.6
32 NO STEP -1.52e-05 4.11e-06 0.000713 1.3
33 NO STEP -3.25e-05 8.4e-06 0.00108 1.6
34 NO STEP -2.93e-05 7.42e-06 0.000994 2.0
35 NO STEP -1.52e-05 4.11e-06 0.000713 2.6
36 29341 0.000673 0.00041 0.000329 2.6
37 NO STEP -4.48e-06 1.01e-06 0.000377 2.2
38 NO STEP -1.02e-05 1.94e-06 0.000507 2.4
39 NO STEP -4.48e-06 1.01e-06 0.000377 2.8
40 NO STEP -6.87e-07 2.39e-07 0.000187 3.4
iteration cost reduction expected gradient log10(λ) η divergence
41 NO STEP -7.03e-08 2.96e-08 5.16e-05 4.2
42 NO STEP -1.2e-09 2.86e-09 8.09e-06 5.2
43 29341 5.78e-07 3.77e-07 7.76e-07 5.2
44 29341 5.64e-07 6e-07 1.24e-06 5.0
45 NO STEP -2.04e-10 1.02e-09 3.16e-06 4.2
46 NO STEP -2.87e-09 2.84e-09 8.08e-06 4.4
47 NO STEP -9.53e-10 1.69e-09 5.06e-06 4.8
48 29341 1.43e-07 9.56e-07 1.98e-06 4.8
49 NO STEP -7.64e-10 1.01e-09 3.16e-06 4.4
50 NO STEP -1.78e-09 1.68e-09 5.05e-06 4.6
iteration cost reduction expected gradient log10(λ) η divergence
51 NO STEP -7.64e-10 1.01e-09 3.16e-06 5.0
52 NO STEP -1.2e-10 3.82e-10 1.23e-06 5.6
53 29341 4e-11 3.99e-10 3.01e-07 5.6
SUCCESS: cost change < tol_fun
iterations: 53
final cost: 29341
final grad: 3.014034e-07
final λ: 2.6959947e+05
time / iter: 609 ms
total time: 32.29 seconds, of which
derivs: 45.3%
back pass: 36.9%
fwd pass: 17.4%
other: 0.5 % (graphics etc.)
=========== end iLQG ===========
Install package Plots.jl (and call using Plots) to plot results in the end of demo_pendcart
Done
==========
Starting box-QP, dimension 500, initial value: 66427.406
RESULT: Gradient norm smaller than tolerance.
iterations 15 gradient 2.39487e-12 final value -31.2095 factorizations 14
2.634966 seconds (1.72 k allocations: 78.097 MiB, 6.56% gc time)
Testing DifferentialDynamicProgramming tests passed
Results with Julia v1.3.1-pre-7704df0a5a
Testing was successful .
Last evaluation was ago and took 2 minutes, 59 seconds.
Click here to download the log file.
Click here to show the log contents.
Resolving package versions...
Installed RecipesBase ──────────────────── v0.7.0
Installed LinearTimeVaryingModelsBase ──── v0.2.1
Installed Requires ─────────────────────── v0.5.2
Installed DifferentialDynamicProgramming ─ v0.4.0
Installed ValueHistories ───────────────── v0.5.1
Installed OrderedCollections ───────────── v1.1.0
Installed Parameters ───────────────────── v0.12.0
Installed DataStructures ───────────────── v0.17.6
Updating `~/.julia/environments/v1.3/Project.toml`
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
Updating `~/.julia/environments/v1.3/Manifest.toml`
[864edb3b] + DataStructures v0.17.6
[9cd584b6] + DifferentialDynamicProgramming v0.4.0
[5e7a79f3] + LinearTimeVaryingModelsBase v0.2.1
[bac558e1] + OrderedCollections v1.1.0
[d96e819e] + Parameters v0.12.0
[3cdcf5f2] + RecipesBase v0.7.0
[ae029012] + Requires v0.5.2
[98cad3c8] + ValueHistories v0.5.1
[2a0f44e3] + Base64
[8ba89e20] + Distributed
[b77e0a4c] + InteractiveUtils
[8f399da3] + Libdl
[37e2e46d] + LinearAlgebra
[56ddb016] + Logging
[d6f4376e] + Markdown
[de0858da] + Printf
[9a3f8284] + Random
[9e88b42a] + Serialization
[6462fe0b] + Sockets
[2f01184e] + SparseArrays
[10745b16] + Statistics
[8dfed614] + Test
[4ec0a83e] + Unicode
Testing DifferentialDynamicProgramming
Status `/tmp/jl_4Ft8gG/Manifest.toml`
[864edb3b] DataStructures v0.17.6
[9cd584b6] DifferentialDynamicProgramming v0.4.0
[5e7a79f3] LinearTimeVaryingModelsBase v0.2.1
[bac558e1] OrderedCollections v1.1.0
[d96e819e] Parameters v0.12.0
[3cdcf5f2] RecipesBase v0.7.0
[ae029012] Requires v0.5.2
[98cad3c8] ValueHistories v0.5.1
[2a0f44e3] Base64 [`@stdlib/Base64`]
[8ba89e20] Distributed [`@stdlib/Distributed`]
[b77e0a4c] InteractiveUtils [`@stdlib/InteractiveUtils`]
[8f399da3] Libdl [`@stdlib/Libdl`]
[37e2e46d] LinearAlgebra [`@stdlib/LinearAlgebra`]
[56ddb016] Logging [`@stdlib/Logging`]
[d6f4376e] Markdown [`@stdlib/Markdown`]
[de0858da] Printf [`@stdlib/Printf`]
[9a3f8284] Random [`@stdlib/Random`]
[9e88b42a] Serialization [`@stdlib/Serialization`]
[6462fe0b] Sockets [`@stdlib/Sockets`]
[2f01184e] SparseArrays [`@stdlib/SparseArrays`]
[10745b16] Statistics [`@stdlib/Statistics`]
[8dfed614] Test [`@stdlib/Test`]
[4ec0a83e] Unicode [`@stdlib/Unicode`]
[ Info: Compile time is high for this package, this is expected and is not an error.
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.1247 17.6 17.6 0.0911 0.0
2 32.5587 12 12 0.0906 -0.2
3 20.5755 8.6 8.6 0.113 -0.6
4 11.9772 4.61 4.61 0.144 -1.2
5 7.36242 1.78 1.78 0.168 -2.0
6 5.58242 0.428 0.428 0.219 -3.1
7 5.15461 0.0223 0.0223 0.0703 -4.3
8 5.13227 2.27e-05 2.27e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.132252
final grad: 3.874959e-06
final λ: 4.4841551e-08
time / iter: 302 ms
total time: 2.72 seconds, of which
derivs: 0.0 %
back pass: 37.5%
fwd pass: 7.5 %
other: 54.9% (graphics etc.)
=========== end iLQG ===========
17.329611 seconds (16.50 M allocations: 991.600 MiB, 4.31% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.0885 20.8 20.8 0.107 0.0
2 29.2398 9.11 9.11 0.0838 -0.2
3 20.1308 4.63 4.63 0.082 -0.6
4 15.5039 3.22 3.22 0.118 -1.2
5 12.2853 1.77 1.77 0.214 -2.0
6 10.5119 0.363 0.363 0.211 -3.1
7 10.1489 0.0144 0.0144 0.0593 -4.3
8 10.1345 1.31e-05 1.31e-05 0.00201 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.13451
final grad: 3.248093e-06
final λ: 4.4841551e-08
time / iter: 106 ms
total time: 0.95 seconds, of which
derivs: 0.2 %
back pass: 65.8%
fwd pass: 29.7%
other: 4.3 % (graphics etc.)
=========== end iLQG ===========
1.006627 seconds (1.35 M allocations: 266.722 MiB, 7.94% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0601 21.5 21.5 0.109 0.0
2 29.5361 10.6 10.6 0.0916 -0.2
3 18.8912 6.07 6.07 0.0966 -0.6
4 12.8247 4.26 4.26 0.134 -1.2
5 8.56329 2.68 2.68 0.251 -2.0
6 5.87883 0.587 0.587 0.254 -3.1
7 5.29149 0.0186 0.0186 0.0679 -4.3
8 5.27286 1.28e-05 1.28e-05 0.00229 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.272849
final grad: 3.49079e-06
final λ: 4.4841551e-08
time / iter: 77 ms
total time: 0.69 seconds, of which
derivs: 0.2 %
back pass: 66.9%
fwd pass: 26.0%
other: 6.9 % (graphics etc.)
=========== end iLQG ===========
0.820973 seconds (1.35 M allocations: 266.722 MiB, 4.06% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.2998 20.4 20.4 0.101 0.0
2 29.8672 10.5 10.5 0.0849 -0.2
3 19.4016 7.13 7.13 0.102 -0.6
4 12.2673 4.63 4.63 0.136 -1.2
5 7.64203 2.03 2.03 0.183 -2.0
6 5.61641 0.353 0.353 0.182 -3.1
7 5.26311 0.0122 0.0122 0.0591 -4.3
8 5.25088 9.73e-06 9.73e-06 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.250866
final grad: 3.203902e-06
final λ: 4.4841551e-08
time / iter: 78 ms
total time: 0.71 seconds, of which
derivs: 0.1 %
back pass: 66.2%
fwd pass: 28.9%
other: 4.7 % (graphics etc.)
=========== end iLQG ===========
0.766293 seconds (1.35 M allocations: 266.722 MiB, 5.11% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.7243 26.2 26.2 0.113 0.0
2 23.4751 8.95 8.95 0.0763 -0.2
3 14.5222 3.04 3.04 0.0648 -0.6
4 11.4817 1.14 1.14 0.0717 -1.2
5 10.3406 0.748 0.748 0.13 -2.0
6 9.59252 0.308 0.308 0.184 -3.1
7 9.28464 0.018 0.018 0.055 -4.3
8 9.26669 1.85e-05 1.85e-05 0.00208 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 9.266675
final grad: 3.309447e-06
final λ: 4.4841551e-08
time / iter: 70 ms
total time: 0.63 seconds, of which
derivs: 0.2 %
back pass: 66.5%
fwd pass: 27.5%
other: 5.8 % (graphics etc.)
=========== end iLQG ===========
0.699264 seconds (1.35 M allocations: 266.722 MiB, 4.55% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.6086 26.8 26.8 0.116 0.0
2 23.799 11.6 11.6 0.0886 -0.2
3 12.1548 4.97 4.97 0.0778 -0.6
4 7.18008 2.4 2.4 0.0833 -1.2
5 4.77537 1.18 1.18 0.135 -2.0
6 3.59532 0.237 0.237 0.167 -3.1
7 3.35866 0.0103 0.0103 0.0556 -4.3
8 3.34832 1.16e-05 1.16e-05 0.00203 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.348307
final grad: 3.361818e-06
final λ: 4.4841551e-08
time / iter: 77 ms
total time: 0.69 seconds, of which
derivs: 0.2 %
back pass: 70.2%
fwd pass: 24.4%
other: 5.2 % (graphics etc.)
=========== end iLQG ===========
0.810153 seconds (1.35 M allocations: 266.722 MiB, 8.40% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 48.8912 22 22 0.103 0.0
2 26.9357 11.1 11.1 0.0851 -0.2
3 15.8772 6.43 6.43 0.0924 -0.6
4 9.45055 3.98 3.98 0.119 -1.2
5 5.46989 2.52 2.52 0.172 -2.0
6 2.95429 0.949 0.949 0.205 -3.1
7 2.0054 0.0605 0.0605 0.0776 -4.3
8 1.94491 6.9e-05 6.9e-05 0.00298 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 1.944839
final grad: 4.641227e-06
final λ: 4.4841551e-08
time / iter: 67 ms
total time: 0.60 seconds, of which
derivs: 0.2 %
back pass: 63.3%
fwd pass: 29.4%
other: 7.1 % (graphics etc.)
=========== end iLQG ===========
0.715190 seconds (1.35 M allocations: 266.722 MiB, 4.19% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.6099 26.7 26.7 0.106 0.0
2 24.8642 11.5 11.5 0.0823 -0.2
3 13.3665 4.88 4.88 0.076 -0.6
4 8.48701 2.25 2.25 0.0841 -1.2
5 6.23922 0.851 0.851 0.143 -2.0
6 5.38792 0.111 0.111 0.132 -3.1
7 5.27669 0.00519 0.00519 0.0456 -4.3
8 5.27149 6.93e-06 6.93e-06 0.00189 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 5.271487
final grad: 3.213114e-06
final λ: 4.4841551e-08
time / iter: 68 ms
total time: 0.61 seconds, of which
derivs: 0.2 %
back pass: 63.3%
fwd pass: 30.5%
other: 6.0 % (graphics etc.)
=========== end iLQG ===========
0.698113 seconds (1.35 M allocations: 266.722 MiB, 4.76% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 51.0102 26.2 26.2 0.115 0.0
2 24.8497 10.8 10.8 0.085 -0.2
3 14.0186 5.39 5.39 0.0863 -0.6
4 8.62521 3.01 3.01 0.107 -1.2
5 5.61383 1.65 1.65 0.169 -2.0
6 3.96697 0.404 0.404 0.187 -3.1
7 3.56312 0.018 0.018 0.0558 -4.3
8 3.54514 1.79e-05 1.79e-05 0.00216 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 3.545121
final grad: 3.581447e-06
final λ: 4.4841551e-08
time / iter: 69 ms
total time: 0.62 seconds, of which
derivs: 0.2 %
back pass: 63.7%
fwd pass: 29.1%
other: 7.0 % (graphics etc.)
=========== end iLQG ===========
0.725521 seconds (1.35 M allocations: 266.723 MiB, 4.75% gc time)
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 50.642 15.2 15.2 0.0855 0.0
2 35.3962 9.74 9.74 0.0823 -0.2
3 25.6515 7.09 7.09 0.105 -0.6
4 18.5648 4.8 4.8 0.16 -1.2
5 13.7694 2.4 2.4 0.227 -2.0
6 11.372 0.627 0.627 0.233 -3.1
7 10.745 0.0337 0.0337 0.0694 -4.3
8 10.7113 4.03e-05 4.03e-05 0.00245 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 10.71122
final grad: 3.964869e-06
final λ: 4.4841551e-08
time / iter: 75 ms
total time: 0.67 seconds, of which
derivs: 0.1 %
back pass: 71.5%
fwd pass: 26.0%
other: 2.4 % (graphics etc.)
=========== end iLQG ===========
0.762954 seconds (1.35 M allocations: 266.723 MiB, 3.89% gc time)
Running linear demo function for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 49.3316 13.8 13.8 0.0813 0.0
2 35.5392 8.61 8.61 0.0773 -0.2
3 26.9284 6.45 6.45 0.099 -0.6
4 20.4758 4.1 4.1 0.149 -1.2
5 16.3735 1.5 1.5 0.196 -2.0
6 14.8783 0.223 0.223 0.163 -3.1
7 14.6555 0.0077 0.0077 0.0513 -4.3
8 14.6478 7.76e-06 7.76e-06 0.00194 -5.7
SUCCESS: gradient norm < tol_grad
iterations: 9
final cost: 14.64778
final grad: 3.237175e-06
final λ: 4.4841551e-08
time / iter: 74 ms
total time: 0.66 seconds, of which
derivs: 0.2 %
back pass: 68.9%
fwd pass: 25.7%
other: 5.3 % (graphics etc.)
=========== end iLQG ===========
5.519023 seconds (3.12 M allocations: 353.599 MiB, 2.03% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Running linear demo function with KL-divergence constraint for DifferentialDynamicProgramming.jl
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 22.3322 26 17.3 0.123 -1.00 0.0095 1.83
2 1.28069e+10 -1.28e+10 5.37e+18 1.41e+07 -0.15 1.28e+10 -1.42
3 4.42424e+07 -4.42e+07 2.1e+15 5.75e+05 -0.07 4.39e+07 0.248
4 225219 -2.25e+05 9.57e+11 1.95e+04 -0.04 2.17e+05 1.07
5 1184.11 -1.14e+03 1.26e+09 945 -0.02 975 1.47
6 10.5919 37.7 7.69e+06 78.6 -0.03 1.46 1.67
7 123.307 -75 1.29e+08 312 -0.03 80.1 1.57
8 417.204 -369 4.22e+08 557 -0.03 317 1.52
9 232.39 -184 2.36e+08 420 -0.03 166 1.55
10 170.332 -122 1.75e+08 363 -0.03 116 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 145.148 -96.8 1.5e+08 337 -0.03 96.9 1.57
SUCCESS: abs(KL-divergence) < kl_step
iterations: 11
final cost: 145.1481
final grad: 336.8931
final λ: 9.3750325e-01
time / iter: 306 ms
total time: 3.37 seconds, of which
derivs: 0.0 %
back pass: 67.0%
fwd pass: 22.5%
other: 10.5% (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 145.1481073547954
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 145.037 0.111 0.0553 0.000151 -1.00 5.58e-05 1.57
2 2.90811e+55 -2.91e+55 1.05e+60 5.18e+25 -0.15 2.92e+55 -1.16
3 1.15096e+52 -1.15e+52 1.9e+56 2.01e+24 -0.07 1.17e+52 0.501
4 3.37113e+43 -3.37e+43 2.88e+47 2.66e+20 -0.04 4.69e+43 1.28
5 4.71286e+11 -4.71e+11 4.77e+15 2.03e+05 -0.02 4.61e+12 1.55
6 145.033 0.115 0.095 0.00299 -0.03 9.61e-05 1.57
7 145.657 -0.509 584 0.574 -0.03 0.564 1.56
8 12564.7 -1.24e+04 1.85e+08 115 -0.03 1.82e+05 1.56
9 138.563 6.59 1.43e+05 6.26 -0.03 142 1.56
10 145.127 0.0214 6.43e+03 1.76 -0.03 6.33 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 143.3 1.85 2.83e+04 3.27 -0.03 28 1.56
EXIT: η > ηmax
iterations: 11
final cost: 143.2996
final grad: 3.267209
final λ: 9.3442150e-01
time / iter: 129 ms
total time: 1.42 seconds, of which
derivs: 0.0 %
back pass: 47.2%
fwd pass: 42.6%
other: 10.2% (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 143.2996378785932
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 143.191 0.108 0.0541 0.000122 -1.00 5.46e-05 1.56
2 1.55674e+56 -1.56e+56 7.17e+60 1.68e+26 -0.15 1.57e+56 -1.15
3 1.68786e+52 -1.69e+52 2.71e+56 2.71e+24 -0.07 1.72e+52 0.505
4 3.1857e+43 -3.19e+43 2.94e+47 2.96e+20 -0.04 4.49e+43 1.28
5 1.39165e+11 -1.39e+11 9.75e+14 8.81e+04 -0.02 9.38e+11 1.55
6 143.187 0.113 0.0787 0.00117 -0.03 7.96e-05 1.56
7 143.893 -0.593 27.3 0.12 -0.03 0.0269 1.56
8 1666.27 -1.52e+03 2.06e+07 39.2 -0.03 2.03e+04 1.55
9 143.946 -0.647 8e+03 1.58 -0.03 7.93 1.56
10 145.28 -1.98 3.64e+05 7.52 -0.03 359 1.56
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 142.342 0.958 5.21e+04 3.4 -0.03 51.6 1.56
EXIT: η > ηmax
iterations: 11
final cost: 142.3417
final grad: 3.39533
final λ: 9.3196340e-01
time / iter: 147 ms
total time: 1.62 seconds, of which
derivs: 0.0 %
back pass: 50.2%
fwd pass: 44.0%
other: 5.8 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 142.3416956113604
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 142.237 0.104 0.0522 0.000152 -1.00 5.26e-05 1.56
2 4.33473e+55 -4.33e+55 2.03e+60 1.08e+26 -0.15 4.37e+55 -1.15
3 2.29364e+52 -2.29e+52 3.65e+56 4.75e+24 -0.07 2.33e+52 0.507
4 2.92665e+43 -2.93e+43 2.9e+47 3.69e+20 -0.04 4.18e+43 1.28
5 4.61146e+10 -4.61e+10 2.91e+14 4.56e+04 -0.02 2.79e+11 1.55
6 142.23 0.112 0.0977 0.0015 -0.03 9.85e-05 1.56
7 142.844 -0.502 13.9 0.0458 -0.03 0.0138 1.55
8 467.82 -325 3.21e+06 15.5 -0.03 3.18e+03 1.55
9 144.594 -2.25 813 0.521 -0.03 0.807 1.55
10 150.678 -8.34 4.06e+04 2.68 -0.03 40.2 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 179.865 -37.5 3.5e+05 6.39 -0.03 347 1.55
EXIT: η > ηmax
┌ Warning: KL divergence too high for some time steps when done
└ @ DifferentialDynamicProgramming ~/.julia/packages/DifferentialDynamicProgramming/DG1NV/src/iLQGkl.jl:248
iterations: 11
final cost: 179.865
final grad: 6.38689
final λ: 9.3073677e-01
time / iter: 115 ms
total time: 1.27 seconds, of which
derivs: 0.0 %
back pass: 49.4%
fwd pass: 45.2%
other: 5.5 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 179.8649740066802
---------- begin iLQG ----------
iteration est. cost reduction expected gradient log10(η) divergence entropy
1 179.749 0.116 0.0578 0.000165 -1.00 5.82e-05 1.55
2 4.79408e+56 -4.79e+56 1.98e+61 2.05e+26 -0.15 4.84e+56 -1.15
3 3.109e+52 -3.11e+52 4.99e+56 4.39e+24 -0.07 3.17e+52 0.509
4 2.60718e+43 -2.61e+43 2.8e+47 3.42e+20 -0.04 3.77e+43 1.29
5 1.48368e+10 -1.48e+10 9.3e+13 2.28e+04 -0.02 8.9e+10 1.54
6 179.709 0.156 0.33 0.00264 -0.03 0.00033 1.55
7 179.366 0.499 55.2 0.0431 -0.03 0.0549 1.55
8 363.425 -184 7.17e+05 5.69 -0.03 708 1.55
9 180.244 -0.379 530 0.222 -0.03 0.526 1.55
10 190.819 -11 8.04e+03 0.95 -0.03 7.96 1.55
iteration est. cost reduction expected gradient log10(η) divergence entropy
11 220.85 -41 7.1e+04 2.28 -0.03 70.2 1.55
EXIT: η > ηmax
iterations: 11
final cost: 220.8503
final grad: 2.281649
final λ: 9.2951176e-01
time / iter: 115 ms
total time: 1.26 seconds, of which
derivs: 0.0 %
back pass: 51.2%
fwd pass: 43.8%
other: 5.0 % (graphics etc.)
=========== end iLQG ===========
Outer loop: Cost = 220.85032087779095
18.132071 seconds (22.29 M allocations: 3.500 GiB, 5.61% gc time)
Install package Plots.jl (and call using Plots) to plot results in the end of demo_linear
Entering iLQG function
---------- begin iLQG ----------
iteration cost reduction expected gradient log10(λ) η divergence
1 33409.7 2.21e+03 3.57e+03 2.35 0.0
2 31195 705 1.57e+03 0.7 -0.2
3 30489.9 537 1.82e+03 0.719 -0.6
4 29952.6 13.9 479 0.406 -1.2
5 29938.7 224 896 0.427 -2.0
6 29714.5 257 260 0.254 -3.1
7 29457.7 10.2 86.3 0.159 -4.3
8 29447.5 79.8 90.3 0.16 -5.7
9 29367.7 18.1 20.4 0.0844 -7.3
10 29349.5 4.78 5.23 0.0469 -9.2
iteration cost reduction expected gradient log10(λ) η divergence
11 29344.8 0.633 1.81 0.027 -11.2
12 29344.1 1.25 1.62 0.0218 -13.5
13 29342.9 1.38 2.12 0.0204 -15.9
14 29341.5 0.221 0.471 0.0109 -18.6
15 29341.3 0.0322 0.111 0.00601 -21.4
16 29341.2 0.0121 0.0284 0.00338 -24.5
17 29341.2 0.0289 0.00881 0.00194 -27.8
18 29341.2 0.146 0.00967 0.00162 -31.2
19 NO STEP -8.15e-05 4.49e-05 0.00151 -6.0
20 NO STEP -8.15e-05 4.49e-05 0.00151 -5.8
iteration cost reduction expected gradient log10(λ) η divergence
21 NO STEP -8.15e-05 4.49e-05 0.00151 -5.4
22 NO STEP -8.15e-05 4.49e-05 0.00151 -4.8
23 NO STEP -8.15e-05 4.49e-05 0.00151 -4.0
24 NO STEP -8.15e-05 4.49e-05 0.00151 -2.9
25 NO STEP -8.15e-05 4.49e-05 0.00151 -1.7
26 NO STEP -8.15e-05 4.49e-05 0.00151 -0.3
27 NO STEP -8.1e-05 4.49e-05 0.00151 1.3
28 NO STEP -6.37e-05 4.28e-05 0.00142 3.2
29 29341 0.0181 0.00494 0.000294 3.2
30 29341 0.0108 0.00163 0.000587 3.0
iteration cost reduction expected gradient log10(λ) η divergence
31 29341 0.0124 0.00142 0.000705 2.6
32 NO STEP -1.52e-05 4.11e-06 0.000713 1.3
33 NO STEP -3.25e-05 8.4e-06 0.00108 1.6
34 NO STEP -2.93e-05 7.42e-06 0.000994 2.0
35 NO STEP -1.52e-05 4.11e-06 0.000713 2.6
36 29341 0.000673 0.00041 0.000329 2.6
37 NO STEP -4.48e-06 1.01e-06 0.000377 2.2
38 NO STEP -1.02e-05 1.94e-06 0.000507 2.4
39 NO STEP -4.48e-06 1.01e-06 0.000377 2.8
40 NO STEP -6.87e-07 2.39e-07 0.000187 3.4
iteration cost reduction expected gradient log10(λ) η divergence
41 NO STEP -7.03e-08 2.96e-08 5.16e-05 4.2
42 NO STEP -1.2e-09 2.86e-09 8.09e-06 5.2
43 29341 5.78e-07 3.77e-07 7.76e-07 5.2
44 29341 5.64e-07 6e-07 1.24e-06 5.0
45 NO STEP -2.04e-10 1.02e-09 3.16e-06 4.2
46 NO STEP -2.87e-09 2.84e-09 8.08e-06 4.4
47 NO STEP -9.53e-10 1.69e-09 5.06e-06 4.8
48 29341 1.43e-07 9.56e-07 1.98e-06 4.8
49 NO STEP -7.64e-10 1.01e-09 3.16e-06 4.4
50 NO STEP -1.78e-09 1.68e-09 5.05e-06 4.6
iteration cost reduction expected gradient log10(λ) η divergence
51 NO STEP -7.64e-10 1.01e-09 3.16e-06 5.0
52 NO STEP -1.2e-10 3.82e-10 1.23e-06 5.6
53 29341 4e-11 3.99e-10 3.01e-07 5.6
SUCCESS: cost change < tol_fun
iterations: 53
final cost: 29341
final grad: 3.014034e-07
final λ: 2.6959947e+05
time / iter: 560 ms
total time: 29.67 seconds, of which
derivs: 44.1%
back pass: 38.1%
fwd pass: 17.2%
other: 0.5 % (graphics etc.)
=========== end iLQG ===========
Install package Plots.jl (and call using Plots) to plot results in the end of demo_pendcart
Done
==========
Starting box-QP, dimension 500, initial value: 66427.406
RESULT: Gradient norm smaller than tolerance.
iterations 15 gradient 2.39487e-12 final value -31.2095 factorizations 14
1.574867 seconds (1.72 k allocations: 78.097 MiB, 4.52% gc time)
Testing DifferentialDynamicProgramming tests passed